make_grounder

make_grounder(grounder_hint: Iterable[LiteralMapping] | Grounder[R], *, implementation: Literal['gilda'] | None = None, progress: bool = False, **kwargs: Any) → Grounder[R][source]
make_grounder(grounder_hint: str | Path | gilda.Grounder, *, implementation: Literal['gilda'] | None = None, progress: bool = False, **kwargs: Any) → Grounder[NamableReference]

Get a grounder from literal mappings.

Parameters:

  • grounder_hint –

    An object that can be coerced into a SSSLM-backed grounder. Can be one of the following:

    1. A URL or file path

    2. An iterable of literal mappings

    3. A pre-instantiated grounder or gilda grounder

  • implementation – If literal mappings are passed, what kind of grounder to use

  • progress – If True, show a progress bar when loading literal mappings

  • kwargs – If literal mappings are passed, keyword arguments passed to the construction of the grounder

Returns:

A SSSLM standard grounder

A grounder can be constructed from a URL. In the following example, a pre-processed lexical index of anatomical terms from UBERON, BTO, MeSH, and other resources is loaded from the biolexica project.

import ssslm

url = f"https://github.com/biopragmatics/biolexica/raw/main/lexica/anatomy/anatomy.ssslm.tsv.gz"
grounder = ssslm.make_grounder(url)

match = grounder.get_best_match("purkinje cell")

A grounder can be constructed from literal mappings that are already stored in a Python object. This example uses the same lexical index as above, first loading it by URL.

import ssslm

url = f"https://github.com/biopragmatics/biolexica/raw/main/lexica/anatomy/anatomy.ssslm.tsv.gz"
literal_mappings = ssslm.read_literal_mappings(url)
grounder = ssslm.make_grounder(literal_mappings)

match = grounder.get_best_match("purkinje cell")

A grounder can be constructed from a pre-existing gilda.Grounder object. As SSSLM is extended, this will incorporate other grounder interfaces.

import ssslm
from gilda.api import grounder as gilda_default_grounder

grounder = ssslm.make_grounder(gilda_default_grounder)

match = grounder.get_best_match("purkinje cell")